High-energy radiations such as X-rays, gamma rays and neutrons are commonly used for non-destructive evaluation (NDE) of the internal defects of an object or for examination of the anomalies inside a human body. Radiographic images for either industrial NDE or medical applications can be obtained by radiography-on-film, fluoroscopy, and tomography methods. Each method has its advantages and disadvantages for a specific application.
Film radiography involves producing a sharp, natural size, permanent image of the internal features (e.g. flaws or anomalies) in an object. Such an image is usually not difficult to interpret. However, film radiography is often relatively slow and expensive.
Fluoroscopy or radioscopy entails the conversion of X-ray intensities into light intensities by utilizing a fluorescent screen. By placing the screen in the X-ray beam behind the specimen, one can produce an image of the specimen on the screen. The high X-ray absorbing capability of selected materials could result in low brightness images and hence poor sensitivity. One method to improve the fluoroscopic performance is to use a closed-circuit television (CCTV) camera to transfer the image on the fluorescent conversion screen on to a display monitor, relying on the electronic circuitry to enhance the signal and produce a bright image. Another technique is to use an image intensifier tube to convert X-rays into photons, which are then picked up by an image sensor. Commonly used image sensors are tube type TV cameras such as isocon, vidicons, and solid state charge coupled device (CCD) cameras. Another type of image sensor is the linear diode array (LDA), which can digitize and store the image to be viewd on a TV monitor. The digitization of the television signal has allowed a computer to be built into the system, and this advancement has greatly improved the attainable image quality. This development has also made it possible to perform real-time radiography.
Both the conventional film radiography and fluoroscopy only provide a two-dimensional (2-D) view of an object. In industrial applications, a 2-D image does not give a NDE technician an adequate perspective view on the spatial distribution of multiple flaws in a structural component, nor does it allow the technician to determine the depth of a particular flaw. For medical uses, a 2-D image may not provide a diagnostician adequate information as to the extent of a particular disorder, such as the exact depth of a foreign object in a human body.
To overcome some of the drawbacks of 2-D radiography, the approach of tomography was developed. Computed tomography (CT) involves obtaining and stacking a sequence of images representing 2-D cross sections or "slices" of the object. The 2-D images are acquired by rotating a thin, fan shaped beam of X-ray about the long axis of the object. X-ray attenuation measurements are obtained from many different directions across each slice. The 2-D images are reconstructed from these data through a sophisticated mathematical convolution and back projection procedure. A major drawback of tomography is that a NDE technician or diagnostician must mentally "stack" an entire series of 2-D slices in order to infer the structure of a 3-D object. The interpretation of a series of stacked 2-D images by an observer requires a great deal of specialized knowledge and skill. Further, such an approach is extremely time consuming and is prone to inaccuracy. The market price of a CT system typically exceeds a million U.S. dollars and, therefore, only select large hospitals or highly specialized governmental or industrial facilities could afford to have a CT system. Clearly, a need exists to develop a more affordable stereography system for 3-D inspection of the internal structure of an object.
Three-dimensional (3-D) or stereoscopic viewing provides a means for showing actual, more understandable spatial relationships among various features or flaws inside a body. Stereoscopic radiology was first introduced near the turn of the century, e.g. L. W. Pease, U.S. Pat. No. 1,447,399 (1923). Extensive patent and open literature can be found that describes the methods or apparatus for producing stereoscopic radiographs. U.S. Pat. No. 5,233,639 (1993), issued to Marks, summarized the pros and cons of various stereoscopic radiography methods, including stereo film radiography and stereoscopic fluoroscopy.
Most of the techniques that have been used to achieve the stereo effect is based on the theory of parallax. Specifically, an image recorded from the perspective of the right eye must be seen by the right eye while an image recorded from the perspective of the left eye must be seen by the left eye. A simple way to accomplish this is to provide distinct and separate optical paths to each eye from each recorded image. For instance, the right and left eye image pairs may be recorded as transparencies which, when inserted in a common hand-held 3-D viewer, are presented to each eye separately through magnifying lenses. A second example using the principle of distinct and separate optical paths is the mirror based viewer system. In this system, the image pairs are positioned under a viewer which, through two pairs of angled mirrors, directs each image to its corresponding observing eye. These conventional 3-D viewers, normally without proper markers or references, do provide the observer a 3-D perspective. However, they do not readily permit determination of the specific depths in which certain features (or flaws) are located relative to a predetermined reference.
Disclosed in U.S. Pat. No. 3,984,684 (1976) is a technique that allows both production of the stereo effect and measurements of the depth and size of one or more internal parts of an object. The technique entails successively directing the X-ray beams from an X-ray tube through the object, then through a parallax grating, and finally onto the film. The grating is mounted on the film support system. The object and the film support system together are translated in parallel paths laterally with respect to the beam path at different speeds. These speeds are such that the film and the object are maintained in congruent alignment with the X-ray tube. The grating moves slightly out of congruency causing the beam passing through the grating to slightly scan the film during the transverse. Also, the angle at which the object is exposed to radiation from the X-ray tube gradually changes. The film image contains a series of side-by-side variable aspect views or images of the object, corresponding in number to the number of slits in the grating. These images when viewed with a lenticular screen produce a 3-D perception. This technique requires the utilization of a complicated radiograph-taking system and a lenticular screen as described above. The stringent congruent alignment requirement has made this technique not readily adaptable to existing X-ray radiography apparatus.
Liu and co-workers (International Journal of Pressure Vessels & Piping, Vol.44, 1990, pp.353-364 and Vol.48, 1991, pp.331-341) have proposed a quantitative stereoscopic method which not only provides a 3-D perspective view of the internal features but permits convenient calculations of the coordinates (X,Y,Z) of one or more flaws inside an object. The method begins with taking a pair of radiograph films with the X-ray tube shifted laterally in a plane parallel to the film between the two exposures (while the object remains stationary). Alternatively, the same result can be achieved by shifting the object laterally while the X-ray source remains fixed. These radiograph films are then examined in a stereoscopic viewer. With a suitable marker placed on the specimen surface when the radiographic films are being exposed, the position of a defect image inside the specimen can be determined. Hitherto, very few industrial stereo radiographs have yielded very good results possibly because of the lack of reference detail and the incapability of the conventional stereoscopes in coping with films of the density used in industrial radiology. The method proposed by Liu, et al. provides a sound basis upon which more effective stereoscopes for quantitative radiography can be designed. This method, however, has been limited to film radiography. What is clearly needed is an improved method, which is based on Liu's principle and the various positive attributes of fluoroscopy, for conducting quantitative stereo radiology. The present invention has come about by following this line of thinking to develop methods for quantitative stereoscopic radiology.